基于支持向量机的TM遥感影像土地利用/覆盖类型监测  

TM Remote Sensing Images Land Use / Cover Monitoring Based on Support Vector Machine

作　　者：荆鑫[1,2] 舒清态[1] 刘庆生[2] 

机构地区：[1]西南林业大学,云南昆明650224 [2]中国科学院地理科学与资源研究所,北京100101

出　　处：《安徽农业科学》2014年第22期7631-7632,7635,共3页Journal of Anhui Agricultural Sciences

基　　金：中国科学院战略性先导科技专项资助项目(XDA05050601);国家自然科学基金重点项目(41023010)

摘　　要：基于TM遥感数据和林业二调数据,对云南省宜良县北古城镇进行土地监测和土地利用分类,利用ArcGIS和ENVI软件对原始影像数据进行裁剪,运用支持向量机(SVM)监督分类方法,将研究区的土地利用类型分为农地、建筑用地、水域、针叶林、裸地6个类别,并且达到一定的分类精度,以图像的形式直观地展示研究区的土地覆盖类型,并以林业二调数据为ROI(感兴趣区域)来验证分类精度。在此基础上,指出了研究区土地利用中存在的问题和生态结构的变化,并提出了调整对策。Based on TM remote sensing data and forestry second survey,land monitoring and land use classification was carried out in north ancient town,Yiliang County,Yunnan Province.In ARCGIS and ENVI,the cutting of the original image data was conducted.Using support vector machine （SVM） supervised classification method,the land use types in the study area was classified into agricultural land,construction land,water,coniferous forest,bare and to a certain classification accuracy.The land cover types were displayed in the form of image,and forestry two tone data ROI,to verify the classification accuracy.On this basis,existing problems in land use and variation of ecological structure were pointed out,the adjustment countermeasures were put forward.

关 键 词：TM 土地利用 遥感分类 林业二调 支持向量机 

分 类 号：S28[农业科学—农业水土工程]